Ultra-Low-Power IoT: Achieving <1µA Sleep Current

A practical guide to designing battery-powered IoT devices that run for years on a single coin cell.

Introduction

Power consumption is arguably the most critical factor in IoT device design. A device that drains its battery in weeks is a maintenance nightmare; one that runs for years is a viable commercial product. Achieving ultra-low power consumption requires understanding the entire system—not just the microcontroller, but every component, sensor, and radio.

This article explores the principles, techniques, and real-world examples of designing ultra-low-power IoT devices. Whether you’re building a soil moisture sensor for agriculture, a temperature logger for cold chain monitoring, or a wearables health tracker, the strategies outlined here will help you maximize battery life.

1. The Consumption Challenge

1.1 Why Power Matters

Battery-powered IoT devices are ubiquitous in real-world deployments. Whether deployed in remote agricultural fields, smart buildings, or industrial facilities, these devices must operate for months or years without battery changes. The cost of battery replacement often exceeds the cost of the device itself—making long battery life a critical business requirement.

1.2 Key Insight: It’s Not Just the MCU

A common mistake is focusing solely on the microcontroller’s power consumption. While important, the MCU is often not the largest power consumer. Consider:

Sensors may draw significant current during measurement
Radios can consume 10-100x more than the MCU during transmission
Voltage regulators can waste power as heat
Idle peripherals may continue consuming power if not properly managed

Example: A sensor that draws 5mA during a 100ms measurement once per minute is acceptable. But the same sensor drawing 1mA continuously will drain a 2000mAh battery in 83 days—regardless of the MCU’s sleep current.

1.3 The Power Budget Equation

The total energy consumed by a device can be calculated as:

Total Energy = (Idle Current × Idle Time) + (Active Current × Active Time)

For ultra-low-power operation, the goal is to maximize idle time and minimize both idle and active currents. This is achieved through:

Deep sleep modes with minimal current draw
Duty cycling—waking up only when needed
Optimising active time—doing work as quickly as possible
Powering off peripherals when not in use

2. Components of Power Consumption

2.1 Microcontroller Power

Controller	Active Current	Sleep Current	Wake-up Time	Best Use Case
STM32L0	~5 mA @ 32 MHz	<2 µA	~5 µs	Battery sensors, 2+ years on coin cell
STM32L4	~8 mA @ 80 MHz	<1 µA	~3 µs	Battery sensors + moderate processing
STM32U5	~6 mA @ 160 MHz	<0.9 µA	~2 µs	Best-in-class low power + performance
ESP32	~100-150 mA @ 240 MHz	10-150 µA	~1-3 ms	Low-power Wi-Fi (deep sleep)
ESP8266	~70 mA @ 80 MHz	10-20 µA	~1-2 ms	Simple Wi-Fi sensors
RP2040	~20 mA @ 133 MHz	~100-200 µA	~1 ms	Cost-sensitive, moderate power
nRF52840	~6-10 mA @ 64 MHz	<2 µA	~3 µs	BLE applications, wearables

Key takeaway: For battery-powered applications, STM32L0/L4/U5 and nRF52 series are the top choices. ESP32 can be acceptable with aggressive deep sleep (e.g., waking every 15 minutes).

2.2 Sensor Power

Sensors vary widely in power consumption. Critical factors:

Measurement current—current drawn during a reading
Measurement time—how long the sensor needs to stabilise
Idle/standby current—current when not measuring
Power-up time—time to become ready after power-on

Sensor Type	Measurement Current	Idle Current	Measurement Time
DHT22 (humidity/temp)	1-2.5 mA	0.1-1 µA	1-2 seconds
BME280 (temp/humidity/pressure)	0.3-3.6 µA	<0.1 µA	<10 ms
SHT30 (humidity/temp)	0.5-1.5 µA	<0.2 µA	<10 ms
MPU6050 (accelerometer)	3.9 mA	10 µA	~100 µs
ADXL345 (accelerometer)	23-140 µA	0.1 µA	~100 µs
Max30102 (heart rate/SpO2)	5-10 mA	<1 µA	10-30 ms
TMP102 (temperature)	10 µA	0.5 µA	<100 ms

Key takeaway: Choose sensors with low idle current and fast measurement times. Modern digital sensors (I²C/SPI) are generally better than analog sensors requiring external ADCs.

2.3 Radio Power

Radios are often the largest power consumer in IoT devices. Transmission current can be 10-100x higher than MCU active current.

Radio/Protocol	TX Current	RX Current	Sleep Current	Typical Duty Cycle
LoRa (SX1276)	100-120 mA @ 14dBm	10 mA	2 µA	<0.1% (1 sec every 15 min)
LoRa (SX1276)	50-60 mA @ 7dBm	10 mA	2 µA	<0.1%
BLE (nRF52840)	5-15 mA	4-10 mA	2 µA	1-5% (depending on advertising)
Zigbee (CC2530)	30-40 mA	20 mA	0.5 µA	<1%
Wi-Fi (ESP32)	100-200 mA	80-150 mA	10-150 µA	0.1-1% (short bursts)
NB-IoT (module)	200-300 mA	50-100 mA	5-10 µA	0.01-0.1%

Key takeaway: LoRa, BLE, and Zigbee are orders of magnitude more power-efficient than Wi-Fi for infrequent, small data payloads.

3. The Sensor Problem

3.1 Not All Sensors Support Low Power

This is a critical and often overlooked issue. Many sensors, especially older analog sensors, draw continuous current regardless of whether they are being read. This makes them unsuitable for battery-powered applications.

High-power sensor types to avoid:

Legacy analog sensors (e.g., LM35 temperature sensor) drawing 50-100 µA continuously
Some gas sensors (e.g., MQ-2) requiring preheating at 100-200 mA
Sensors without standby or power-down modes

Low-power alternatives:

Digital sensors with integrated ADC and shutdown modes (e.g., BME280, SHT30)
Sensors with configurable sampling rates (e.g., ADXL345 with wake-up mode)
Sensors that can be completely powered down via a MOSFET

3.2 Powering Down Sensors

A common technique for sensors without low-power modes is to power them through a MOSFET controlled by the MCU. This allows you to completely cut power to the sensor when not in use.

// Example: Powering down a sensor via MOSFET
void read_sensor(void) {
    // Enable power to sensor
    gpio_set_level(POWER_ENABLE_PIN, 1);
    delay_ms(10); // Wait for sensor to stabilise
    
    // Read sensor
    uint16_t value = i2c_read_sensor();
    
    // Power down sensor
    gpio_set_level(POWER_ENABLE_PIN, 0);
}

Advantages:

Sensor draws zero current when not measuring
Works with any sensor, regardless of built-in low-power features

Disadvantages:

Requires an extra MOSFET and GPIO pin
Adds wake-up time (usually negligible)

3.3 Sensor Selection Guidelines

When selecting sensors for battery-powered devices:

Check idle current—should be <1 µA ideally
Check measurement time—shorter is better (<100 ms)
**Look for “low power” or “ultra-low power” in the datasheet
Verify power-down modes—some sensors have no shutdown capability
Consider self-discharge—some sensors (e.g., gas sensors) have inherent power consumption

4. Real-World Example: STM32 Ultra-Low Power

4.1 STM32L0/L4/U5 Series Overview

The STM32L series is the gold standard for ultra-low-power microcontrollers. Key features include:

Multiple sleep modes—from Stop to Standby (<2 µA)
Wake-up from standby via GPIO, RTC, or external events
Ultra-low-power analog peripherals—LPUART, LPTIM, LPDAC
Dynamic voltage scaling—adjust core voltage for performance
Battery backup domain—retains critical data during deep sleep

4.2 Power Modes Comparison

Mode	Current	Wake-up Source	Wake-up Time	Use Case
Run	100 µA/MHz	N/A	N/A	Active processing
Sleep	50-80 µA/MHz	Any interrupt	2-3 cycles	Light sleep with fast wake-up
Low-power Run	25 µA/MHz	N/A	N/A	Continuous low-power processing
Low-power Sleep	10-20 µA	Any interrupt	2-3 cycles	Low-power waiting
Stop 0	~100 µA	All peripherals	~2 µs	Fast wake-up, peripherals active
Stop 1	~1-2 µA	Limited peripherals	~3 µs	Moderate wake-up, peripherals limited
Standby	<2 µA	Reset, RTC, wake-up pins	~5 µs	Deep sleep, minimal power
Shutdown	<0.5 µA	Reset, wake-up pins	~50 ms	Deepest sleep, no RAM retention

4.3 Practical STM32 Power Optimisation

Code example: Entering Standby mode with RTC wake-up

// STM32L0/L4 Ultra-low-power setup
void enter_standby(void) {
    // Disable unused peripherals
    __HAL_RCC_CRC_CLK_DISABLE();
    __HAL_RCC_I2C1_CLK_DISABLE();
    __HAL_RCC_USART2_CLK_DISABLE();
    
    // Configure RTC for wake-up in 15 minutes
    RTC_TimeTypeDef sTime = {0};
    RTC_DateTypeDef sDate = {0};
    RTC_AlarmTypeDef sAlarm = {0};
    sAlarm.AlarmTime.Hours = 0;
    sAlarm.AlarmTime.Minutes = 15;
    sAlarm.AlarmTime.SubSeconds = 0;
    sAlarm.AlarmMask = RTC_ALARMMASK_HOURS;
    HAL_RTC_SetAlarm_IT(&hrtc, &sAlarm, RTC_FORMAT_BIN);
    
    // Enter Standby mode
    HAL_PWR_EnterSTANDBYMode();
}

4.4 Application Example: Battery-Powered Sensor

A soil moisture sensor using STM32L0 + LoRa module:

MCU: STM32L053R8
Sensor: Capacitive soil moisture sensor (ADS1115 + probe)
Radio: SX1276 LoRa module
Power: 2x AA batteries (2,000 mAh each)

Operation cycle (every 15 minutes):

MCU wakes from Standby (0.5 µA) due to RTC alarm
MCU powers sensor via MOSFET (10 ms)
MCU reads sensor (100 ms @ 5 mA)
MCU processes data (1 ms @ 100 µA/MHz)
MCU powers LoRa module (1 second @ 120 mA)
MCU transmits data (1 packet, ~50 bytes)
MCU enters Standby mode

Power consumption calculation:

Standby: 0.5 µA × 898.8 seconds = 0.45 mAh per cycle
Active: 100 µA × 0.02 seconds = 0.0006 mAh per cycle
Sensor: 5 mA × 0.1 seconds = 0.14 mAh per cycle
Radio: 120 mA × 1 second = 33 mAh per cycle
Total per cycle: ~33.6 mAh
Battery life: 4,000 mAh / (33.6 mAh × 96 cycles/day) ≈ 1.2 years

Optimisation for 2+ years:

Reduce transmission frequency (every 30 minutes)
Optimise LoRa transmit power (use 10 dBm instead of 14 dBm)
Use duty cycling for sensor (measure less frequently)
Choose a sensor with lower active current

5. Strategies for Ultra-Low Power

5.1 Duty Cycling

Duty cycling is the most important technique for ultra-low-power operation. The device spends most of its time in deep sleep, waking only to perform a task and immediately returning to sleep.

Key parameters:

Sleep duration—longer sleep means lower average power
Wake-up time—how long it takes to become active
Active time—how long tasks take to complete

Formula:

Average Power = (Sleep Current × Sleep Time + Active Current × Active Time) / Total Time

Example:

Sleep current: 1 µA
Active current: 10 mA
Sleep time: 14 minutes 59 seconds
Active time: 1 second
Average power = (1 µA × 899s + 10mA × 1s) / 900s ≈ 12 µA

5.2 Sleep Mode Selection

Choose the lowest power sleep mode that meets your wake-up requirements:

Requirement	Recommended Mode
Need very fast wake-up (<10 µs)	Stop mode
Need RTC wake-up	Standby mode
Need preserve SRAM data	Stop or Standby with RAM retention
Need preserve GPIO states	Standby or Stop
Need external interrupt wake-up	Stop or Standby
Need ADC/DAC after wake-up	Stop mode (peripherals retain)

5.3 Peripheral Power Management

Key principle: Power off anything you don’t need.

Clock gating—disable clocks to unused peripherals
Power domain control—some MCUs have separate power domains
GPIO configuration—set unused pins to analog input to minimise leakage
Voltage regulator—use low-power regulator modes

Example: STM32 peripheral disable:

// Disable unused clocks
__HAL_RCC_CRC_CLK_DISABLE();
__HAL_RCC_DMA1_CLK_DISABLE();
__HAL_RCC_USART1_CLK_DISABLE();

// Set unused GPIOs to analog input
GPIO_InitTypeDef GPIO_InitStruct = {0};
GPIO_InitStruct.Pin = GPIO_PIN_All;
GPIO_InitStruct.Mode = GPIO_MODE_ANALOG;
GPIO_InitStruct.Pull = GPIO_NOPULL;
HAL_GPIO_Init(GPIOA, &GPIO_InitStruct);
HAL_GPIO_Init(GPIOB, &GPIO_InitStruct);
HAL_GPIO_Init(GPIOC, &GPIO_InitStruct);

5.4 Radio Optimisation

Radios are power-hungry. Optimise their usage:

Reduce transmission power—use the minimum required
Reduce packet length—send only essential data
Aggregate data—send multiple readings in one packet
Use acknowledgements sparingly—they add extra transmissions
Consider using lower-layer protocols—raw LoRa without LoRaWAN for reduced overhead

6. Real-World Application Examples

6.1 Soil Moisture Sensor (Agriculture)

MCU: STM32L052R8
Sensor: Capacitive soil moisture (ADS1115)
Radio: LoRaWAN (SX1276)
Power: 2x AA batteries
Cycle: Wake every 15 minutes, measure, transmit
Battery life: 2+ years
Key optimisations: Deep sleep, short radio burst, low sensor current

6.2 Temperature Logger (Cold Chain)

MCU: STM32L432KC
Sensor: TMP102 or DS18B20
Radio: Bluetooth LE (nRF52840) or LoRa
Power: CR2032 coin cell (220 mAh)
Cycle: Wake every 5 minutes, measure, store (transmit hourly)
Battery life: 1+ year on BLE, longer on LoRa
Key optimisations: Aggressive duty cycling, store-and-forward

6.3 Smart Water Meter

MCU: STM32L0/L4
Sensor: Hall effect flow sensor (external trigger)
Radio: NB-IoT (SIM800/Quectel)
Power: 1x AA battery (3,000 mAh) or Li-SOCl2 (3,000 mAh)
Cycle: Wake on flow event, count, transmit daily
Battery life: 5+ years
Key optimisations: Event-driven wake-up, infrequent transmission

6.4 Wearable Health Monitor

MCU: nRF52840 or STM32L4
Sensor: Max30102 (heart rate/SpO2) + ADXL345 (activity)
Radio: BLE (built-in on nRF52840)
Power: 200 mAh LiPo battery
Cycle: Wake every 1 second (accelerometer), measure HR every 10 minutes
Battery life: 2-4 weeks
Key optimisations: Accelerometer uses activity wake-up, HR only when active

7. N3xar’s Approach to Low Power

The N3xar platform is designed to support a wide range of power profiles, from AC-powered industrial devices to ultra-low-power battery sensors.

Supported Device Types:

AC-powered devices: Gateways, servers, industrial controllers
Powered nodes: Sensors and actuators with stable power
Low-power autonomous nodes: Battery-powered sensors with <1 µA sleep current

Key Products:

Ethernet, wired, Wi-Fi & LoRa devices — flexible connectivity options
I²C, motion, water, and contact sensors — diverse sensing capabilities
Electromechanical actuators & valves — controlled actuation
Stepper/servo motor control boards — precision motion control

Power Management Features:

Dynamic sampling rates based on power availability
Adaptive transmission intervals to conserve battery
Sleep-mode device profiles (<1 µA)
Real-time power monitoring and alerts

“Our system supports both wired and wireless communication and uses two node types: low-power autonomous nodes for long-term sensor use, and powered nodes for sensors and actuators.”

8. Design Checklist for Ultra-Low-Power IoT

Use this checklist to ensure your design is optimised for low power:

MCU Selection

Choose MCU with deep sleep current <2 µA
Verify wake-up time is acceptable for your application
Check for low-power peripherals (LPUART, LPTIM)

Sensor Selection

Check idle current (should be <1 µA if possible)
Verify sensor has power-down mode
Consider MOSFET to completely power down sensor
Check measurement time (shorter is better)

Radio Selection

Choose protocol appropriate for your data rate and range
Consider transmission power (use minimum required)
Verify sleep current of radio module
Check if radio can be powered down completely

Power Supply

Choose regulator with low quiescent current
Consider bypassing regulator for sleep mode (direct battery connection)
Select battery type appropriate for your current profile

Firmware Optimisation

Implement duty cycling (maximise sleep time)
Enter deep sleep immediately after tasks
Disable unused peripherals and clocks
Use low-power modes for peripherals
Optimise radio usage (short bursts, aggregating data)

System-Level

Calculate power budget and battery life
Test with real-world usage patterns
Monitor power consumption during development
Plan for battery replacement or recharging

9. Conclusion

Designing ultra-low-power IoT devices requires a holistic approach that considers every component of the system. Key principles:

Focus on the entire system—not just the MCU
Maximise sleep time with aggressive duty cycling
Choose low-power sensors with fast measurement times
Optimise radio usage—it’s often the largest power consumer
Power down peripherals when not in use
Measure and iterate—power consumption varies in practice

“The best low-power device is one that spends 99.9% of its time in deep sleep, wakes up only when needed, and does its work as quickly as possible before returning to sleep.”

With modern MCUs, sensors, and radios, it is entirely possible to achieve battery life measured in years. The techniques and examples in this article provide a roadmap for achieving that goal.